A Sample-Based Color Correction Method for Laparoscopic Images

Color accuracy of endoscopic and laparoscopic images is of paramount importance for diagnosis and therapy during vision-guided minimally invasive surgery (MIS). However, unsatisfactory lighting conditions would increase the color difference of endoscopic and laparoscopic images. Aiming at improving the color accuracy of the images, a color correction matrix (CCM) can be utilized to transform the tristimulus values of the images to the new ones that are closer to the RGB values of standard color charts. But, a CCM is often related to a fixed illuminance, since the imaged tristimulus values of the same object vary under different illumination. Therefore, to improve the applicability of the CCM method under different illuminance conditions, a dynamic CCM method depending on the image luminance is needed. This paper hence presents a sample-based method that adjusts the CCM's implementation and extends its application for diverse illuminance, particularly in the low-illuminance scenario. The principal idea of the presented method is to correct an image's color under its average relative luminance using an interpolated CCM. The interpolated CCM is calculated from dozens of sample images of a color chart under different illuminance with respect to the images' average relative luminance. And the legitimacy of the interpolated CCM is based on an investigation of the correlation between the obtained CCMs. The experiments were performed on a laparoscope with two high-definition (HD) cameras, and the validations of color correction were conducted on the laparoscopic images captured for both standard color charts and abdominal models, demonstrating the effectiveness of the proposed approach.